Spec a safe filename check for uploads
Your task
Your Python service saves each upload to /srv/uploads/<user_id>/, using the filename the browser sent. A coding agent will write is_safe_filename(name), which decides whether that name may be used on disk. A penetration test last year got a file written outside the user's folder.
Write the spec you would hand the agent.
What you know:
- the server runs Linux, and files are served back later for download
- people upload names like
résumé.pdf,Q3 report (final).xlsxand names in Japanese - the team wants the check written with Python's
remodule - the app must still show the original name, even when the stored name differs
Then, in a few sentences: what does "regex to validate a safe filename" get wrong?